Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.


    As a very very new user of MO — and in contrast to the original poster of this discussion — I was attracted to MO precisely because of the restrictions and guidelines on the nature of the questions. In the past I've tended to avoid online math groups in large part because I prefer to ask and answer questions with well-delineated sets of admissible answers, and I've found that unregulated internet discussion, even on technical subjects, can rapidly degenerate to vagueness and sometimes even to youtube-like cruelty and insipidity.

    Nevertheless, it seems to me that, on MO, whereas a number of questions that are soft, vague, subjective, open-ended, or not research-level — "questions of the first kind," let me call them — enjoy stratospheric popularity (e.g. or e.g. or e.g. or e.g. or e.g.), a number of more precise questions that generally conform to the first three standards of the guidelines — "questions of the second kind" — tend to garner far less attention, especially when they are relatively technical (e.g. or e.g. or e.g. or e.g. or e.g.). This is not surprising, of course, but the effect is then amplified, because it appears that once a question has become unpopular, it is more likely to grow more so: after all, it is difficult to answer, comment upon, or vote on a question that you cannot find, and it is difficult to find questions that do not have many answers, comments, or votes. (In fact, the kiss of death for a question might be for it to receive a small number of up-votes and one partial, up-voted answer. Within a month, it might be very difficult to find.)

    I should say that I have nothing in particular against questions of the first kind, and in particular nothing against those I've cited; they are reasonable questions that should be discussed by someone, somewhere. I tend to prefer questions of the second kind, however, and my understanding of the vision of MO is that its aim is precisely to provide a forum where questions of the second kind can find answers. Yet I still find that, even with a judicious use of tagging, there remains a fair amount of time-consuming sifting one must do to locate questions of the second kind.

    I am thus lead to three questions. (1) Does the easy popularity of questions of the first kind cause questions of the second kind to languish? (2) Are there tools (beyond tagging) that users like me can use to sift more effectively through questions of the first kind in order to be more likely to stumble upon interesting questions of the second kind? (3) Is there a way to counteract the negative feedback loop that causes some questions of the second kind to fall into obscurity?

    My take:
    (1) Yes, to some degree, but it's also a problem of the format.
    (2) Not to my knowledge
    (3) I think that it's a problem of the format.

    For (2): on the right of the site, there is a place to chose 'interesting tags' and 'ignored tags'. Ignoring a tag such as 'soft-questions' makes them appear faded on the homepage and easy to ignore visually (so not a perfect solution, but still a nice one).

    My solution to this problem (or really the more general problem that I can't read all of the questions anymore) is to use google reader. I have a folder of roughly 60 user feeds of either people I know or who tend to answer/comment on questions I am interested in, and I mostly read that.

    • CommentAuthorAnweshi
    • CommentTimeJan 16th 2010 edited

    @clarkbar. I myself asked such a question of the "first type", right upon my arrival. But I was educated quickly and stayed away from such types later. However the temptation to ask is always there. I try to suppress the urge, just as within mathematics I try to handwave less and do actual work more.

    The issue you point out is very valid. I would love to see good quality people, and it is the hope of interacting with them that keeps me here. For that it is necessary to enforce some quality. However it has to be done without being too rude, which does happen sometime. It is hard for someone to face the fact that "you are not good enough", or "your questions are not good enough" and nothing is gained by rubbing it in.

    I have an answer for 3 though. Close those questions of first type after a while. However the askers may be known personally known to the moderators, or may have good standing in mathematical community, and thus it is a difficult matter. However, as people gain in reputation above 5k or 10k or so, threads can be closed by the votes of 5 or more people. Again it is a problem if the "wrong" people keep gaining reputation.

    Again, for any such issues, it will be instructive to see what stackoverflow did. However not all that may be applicable here, since MO is a more serious place, filled with serious researchers.


    (3) Yes, we can encourage a culture where people:

    • edit pure math questions a lot: tags, formatting, latex, references, background, etc.;
    • provide their incomplete but meaningful answers;
    • write very clear questions so that it's harder to give a partial answer;
    • don't forget to mark their non-math questions with one of known tags, so that they can be ignored;
    • don't worry too much about reputation and votes;
    • are not ashamed to revisit their old question and repost part of it, if necessary.

    Explanation for the last bullet: I've often seen people receiving a partial answer to a question, usually because the question was really several related questions. The most common response for this situation I've seen so far is to edit the original post and provide details about the progress. However, it's understandable that people who read question don't return to such a question. So one thing I'm trying to do is to encourage posters to create a new, better-formulated question that has everything currently known as a background. I believe this subdivision of a question is a very useful tool.

    (3) I want to add "use bounties" to Ilya's bullets.

    Have bounties actually been helping people get their questions answered? I don't have a strong impression one way or the other.

    • CommentAuthorHarry Gindi
    • CommentTimeJan 16th 2010 edited
    I gave a 100 point bounty on a question and got one relevant and wholly incomplete response. You're better off reposting the question at least once.

    @Brown: Yes, sorry, the interesting/ignored tags feature is what I was trying to say when I referred to "tagging." (Not up to speed on the parlance here, yet!) But your idea about the Google Reader is interesting; thank you.

    @Anweshi: I'm very concerned about rudeness in online interactions. MO seems to be almost completely free of that, thus far. I hope the trend continues!

    @Nikokoshev: I'm afraid I'm not sure I understand. While I heartily agree that one ought not "worry too much about reputation and votes," it's an unavoidable fact that these numerics are fundamental to the basic functioning of this site and in particular to the life-cycle of questions asked here. That isn't necessarily a bad thing, but it does catalyze a kind of systemic bias toward questions of the first kind, which is what I tried to describe.

    If I understand, you're saying that we might correct this bias purely with culture. I would like to agree with that generally, but your suggestions sound as though you wish to move toward a culture in which the burden for this correction is completely transferred to the askers. In other words, in such a culture a user could not expect an answer merely because he/she wrote a clear, precise, interesting question of the second kind; rather, he/she should have to expect to doctor his/her question incessantly, trying to drum up interest in order to make up the fact that his/her question wasn't sufficiently soft, vague, subjective, elementary, or open-ended to garner enough attention to keep it from slipping into obscurity. Do I misunderstand? Isn't that solving the wrong problem?


    @Clark: yes, I'd like to nudge people toward solving these bias with culture. More precisely, though I don't think this is the only instrument we have, I think this should be the first instrument we turn to whenever the problem is about "wrong something" about the site.

    Let me explain why I think so. To do so, let's first define the problem: the distribution of votes / favorites / views / reputation on the site doesn't agree with what we think would be "reasonable" one.

    I think this is a real problem, and it appears everywhere in real life (good example: politics) which is just one the hints to its depth.

    Therefore, if we start talking about it, we might discover we actually have different ideas about what distribution of votes/views/favs/other numbers is reasonable. To simplify things, let's assume we agree that we prefer math questions ("questions of the second kind" in your words) and correct answers to be upvoted.

    Now here you're right in suggesting a technical fix which may change the distribution of votes toward those goals. But you miss the fact that this distribution will continue to be very far from this ideal. Let me give some examples.

    There are several answers to Is it possible to capture a sphere in a knot?; the ones that are not just comments on others have upvotes. The correct answer has 15 upvotes; there are 3 more interesting attempts to answer that guess the result incorrectly: they have 21, 18, 13. Is this "the wisdom of the crowds gone wrong"?

    In other words, the voters seem to give more to half-answers that appear to lead somewhere rather then the completed answer; is that reasonable?

    Take another example: I have two questions posted roughly at the same time, Explanation for E_8’s torsion and Most interesting mathematics mistake?. I took a lot of time writing the first one to make is as interesting as I could (okay, that could be not so apparent from the title). Yet since the "viewed" count of the second one, at 4 thousand, is ten times more than the viewed count of the E_8 question, it's hopeless to expect that the first would have many more upvotes than the second. Indeed, it has a bit less.

    The last problem could be "solved" by prohibiting soft questions, but if we're not talking about such a remedy, I don't think there's any way of making 4 thousand people more interested in "hard math" questions. So the "view" metric will always accurately reflect this fact; we just need to continue posting hard questions without worrying too much about it.

    toward a culture in which the burden for this correction is completely transferred to the askers.

    Not really, all of the culture things be done by others editing the question (perhaps with the exception of Jonas' suggestion about bounties, but even then others can suggest this to the asker). Even the subdivision of the question can be effectively done by a third party; I presume people are not doing it currently precisely because we aren't yet in the culture of asking only concise questions.

    • CommentAuthorHailong Dao
    • CommentTimeJan 17th 2010 edited

    One simple thing to do regarding (3) is to upvote the technical answers more. I can provide 2 examples that I know personally: this by bhargav and this by Matt Emerton.

    Both would certainly take months for someone who is not an expert to find a satisfying answer (in fact, both answerers themselves do not seem to know of many useful reference). They capture what I think is the best of MO: professional, precise, and very friendly. And it surely takes time to write good technical answers, it is comparable to writing a paper I think. Currently, they get 4 and 3 votes respectively, which I think is a bit unfair. Since I can only vote once for them, the only thing I could do is going through similar answers and upvote!

    (I personally have no problem with soft questions, I enjoy reading many of them. In some sense, for the site to prosper, we need them. But we can try to attain a balance with the more serious questions, which is also a big draw to many people visiting the site.)

    EDIT: I do not know why the links do not appear right, sorry for the inconvenience! How can I fix it?

    Part of the trouble is that I don't want to upvote an answer which I am unable to evaluate. For example, Emerton's answer which Hailong Dao links to looks great in every way I can judge: He uses the vocabulary that I know should show up in an answer to this question, the indices on his sequences go in the right directions, his English is fine. And, based on my past interactions with Emerton, I imagine that this is a great answer. On the other hand, if there is some subtle error in his argument, I know that I would not spot it. I am nervous about voting up an answer in this situation.

    I agree with Hailong that, when you can see that an answer is good, you should vote it up, especially if it is the type of answer you want to see more of.
    David, that is a good point. I certainly did not mean we should go out and upvote anything that looks like a good answer.

    EDIT: I do not know why the links do not appear right, sorry for the inconvenience! How can I fix it?

    Don't do <a href="...", use Markdown syntax instead. Same advice for MathOverflow itself :)

    Ilya said:

    "The last problem could be "solved" by prohibiting soft questions, but if we're not talking about such a remedy, I don't think there's any way of making 4 thousand people more interested in "hard math" questions."

    I was one of those four thousand people, and I am very interested in "hard math" questions. However, my interests lie along the lines of probability, mathematical physics, and geometric analysis. A question on the torsion of E_8, no matter how well-written, is too far astray. By definition, the popular questions are those that appeal to everyone on the site, regardless of mathematical specialty.

    @Nikokoshev. I'm sorry; I think I haven't been clear about my concern:

    (A) I don't object to the general idea of correcting these biases with culture. Again, I'm sympathetic to this idea, but I'm seeking creative ideas for how it might be implemented.

    (B) I don't care about the distribution of votes/favorites/views/reputation at all, except insofar as they inhibit users from stumbling on interesting questions of the second kind — which is, as I understand it, the stated purpose of this site.

    (C) I certainly never suggested "a technical fix which may change the distribution of votes toward those goals." I might, however, suggest a technical fix that may make it easier for sticks-in-the-mud like me to stumble on interesting questions of the second kind; it might be as simple as a checkbox that says: "Show community wiki questions."

    (D) No one would expect a question about E_8 to garner as much attention as a general question about mistakes in mathematics, but that isn't the point. I'm not trying to think of ways of getting 4000 people interested in E_8 (at least, not right now!); my worry is that the relative unpopularity of your E_8 question might mean that the users (especially new users) who would enjoy reading it and thinking about it might not be able to discover it, because it may end up buried under a heap of questions of the first kind.

    @Dao. Your questions (and the ensuing discussions) are among those I've enjoyed reading the most. Thank you. I look forward to trying to answer one soon!

    @Tom, the highest voted question on the site is a question about Bott Periodicity that does not have anything to do with physics, probability, or geometric analysis, or any similarly applied subject.

    @Clark: Thanks for the kind words, and the new answer!